Functional Tertiary Lymphoid Structures in Human Gliomas

Group 21

Introduction

Materials and Methods

Materials:

Methods:

  • Loading data

    • read_csv()

  • Cleaning data

    • Replacing “Not_Available” with NA

    • Removing duplicates: distinct()

    • Standardizing names

  • Creating new metadata columns

    • Age_groups

    • TLS_status_bin

  • Normalizing gene expression

    • cpm_normalization

    • quantile_normalization

  • Computing GEP scores

  • Combining normalized counts with metadata

  • Preparing data for PCA/downstream analysis

Methods:

  • Baseline tables

    • TLS_status: IDH_status, Gender, Age, Localisation
    • IDH_status: TLS_status, Gender, Age, Localisation

  • PCA on gene expression

    • PC1–PC2 coloured by TLS_status
    • PC1–PC2 coloured by IDH_status
    • Variance explained by PCs
  • Model 1 – TLS effect on gene expression
    • CPM_log2 ~ TLS_status_bin
    • Output: volcano plot (TLS+ vs TLS−)
  • Model 2 – GEP scoreby TLS status
    • GEP_score ~ TLS_status_bin
  • Model 3 – TLS and clinical features (Fisher tests)
    • TLS_status_bin vs IDH_status
    • TLS_status_bin vs Primary/Recurrent
    • TLS_status_bin vs Gender
Results

Impact of TLS status on transcriptional signatures

  • TLS status does not strongly separate the samples

  • TLS group contains highly distinct outliers

Results

Impact of IDH status on transcriptional signatures

  • Clear and distinct separation of samples by IDH status

  • Outlier subsets exist

Results

TLS effect on gene expression

  • Each point = one gene (TLS+ vs TLS−)
  • Several genes are significantly associated with TLS status (q < 0.01)
  • Some genes higher in TLS+, others in controls
    → TLS status is linked to differences in gene expression
Results

GEP score by TLS status

  • Each point = one tumor sample
  • GEP score (18-gene immune signature)
  • Median GEP score is higher in TLS+ tumors than in controls
Results

Results

TLS and clinical features (Fisher’s tests)

Discussion